Abstract: This paper proposes a method for detecting missing electronic components on printed circuit boards (PCBs) based on IMAQ Vision Assistant. The method first corrects the image of the PCB to be inspected, and then mainly utilizes the template matching function of IMAQ Vision Assistant to detect missing electronic components. An application example is given, and the results show that this method can effectively detect missing electronic components on PCBs. Keywords: IMAQ Vision Assistant; printed circuit board; pattern matching [b][align=center]Inspection of missing electronic component on Printed Circuit Board based on IMAQ CHEN Fei, XU Xian-zhen, LU Ming-li, XIE Qi[/align][/b] Abstract: A method of inspection of missing electronic component on printed circuit board based on IMAQ Vision Assistant is proposed. Firstly, this method calibrates the inspected image of printed circuit board. Then, the pattern matching function of IMAQ Vision Assistant is mainly used to inspect the missing electronic component on printed circuit board. An application example is given, and the results indicate that this method is effective. Keyword: IMAQ Vision Assistant; printed circuit board; pattern matching 1 Introduction Printed Circuit Board (PCB) is an important component of current electronic products, and Surface Mount Technology (SMT) is the main method used to assemble printed circuit components. Traditional electronic components are pin-type components, while surface mount technology (SMD) places surface mount devices (SMD) directly onto a circuit board coated with solder paste, and then uses reflow soldering to fix the components to the surface of the circuit board [1]. Although surface mount technology has become increasingly mature and the accuracy of pick-and-place machines has been continuously improved, there are still many defects in the pick-and-place process. These defects can cause instability of the entire circuit board or render the entire circuit board without any function, resulting in considerable losses for enterprises. In the assembly process of SMT components on printed circuit boards, the defects that are easy to generate can be divided into four parts: missing components, skew, reverse, and poor soldering. Among them, poor soldering includes open circuits or open circuits, bridging or short circuits, and tombstone effect [2]. In the past, the detection of defects in PCB surface components mostly adopted traditional manual visual inspection and electrical inspection. This detection method is not only inaccurate and time-consuming, but also cannot adapt to high-speed assembly line operations. To address this problem, efforts have been made to develop the Automatic Optical Inspection (AOI) system, which has become one of the emerging detection methods due to its high stability. Generally, different defects require different detection methods. Therefore, this article proposes a method using NI's IMAQ Vision Assistant software to detect missing components, such as resistors, capacitors, and ICs, which are among the most commonly used SMT components on printed circuit boards. The article first briefly introduces the image correction function module and template matching function module of the IMAQ Vision Assistant software, and then provides an application example for detecting missing electronic components on PCBs. The results show that this method can effectively detect missing component defects, and it can also detect component skew and reverse defects. 2. Template Matching Function Module of IMAQ Vision Assistant IMAQ Vision is a vision development toolkit built into LabVIEW (Laboratory Virtual Instrument Engineering Workbench), provided by NI. It includes IMAQ Vision Build and IMAQ Vision Assistant. IMAQ Vision is a powerful function library that provides a large number of image processing function modules and functional modules, such as image acquisition, image calibration, image processing, and geometric measurement, and includes a series of MMX optimized functions. Moreover, the automatic code generation function of IMAQ Vision Assistant greatly shortens the development cycle and reduces costs [3]. (1) Use of the template matching function module [4] Template matching is one of the most important image analysis tools. The template matching function in IMAQ Vision Assistant software can provide information such as the location and matching value of the image being detected that matches the standard template image. To use this function module, you need to: (a) Define a template image (i.e., give the standard template image); (b) Set the parameters of the image being detected: (i) Set the maximum number of matching images to be found that match the standard template image. The maximum value of this parameter can be 1000; (ii) Set the minimum matching value (the range of the matching value is [0, 1000], 1000 means a perfect match and 0 means no match): the larger the value is set, the higher the degree of similarity required for matching, and vice versa. However, if the value is set too high, the qualified printed circuit board will be treated as a defective product. If the value is set too low, the defective printed circuit board will be treated as a qualified product. Therefore, the setting of this value must meet the actual requirements; (iii) Set the angle that the matching area in the standard template image and the image to be detected can exist. The range that can be set is [-α, +α], where α is 0° to 180°; in this parameter setting, if the mirror angle is selected, the range of the angle that the matching area in the standard template image and the image to be detected can exist is [-α, +α] or [-(180-α), +(180-α)]. (2) Template matching method and characteristics Traditional template matching often uses cross-correlation function. The definition of its cross-correlation function [4-6] is as shown in equation (2-1): (2-1) Where w(x,y) is the standard template image of size ; f(x,y) is the image to be detected of size . Equation (2-1) is fine for finding a single match, but if it is used in multiple matching, the result is greatly affected by the light (brightness) of the template and the image to be detected. This method is not only slow, but also has very strict restrictions on the size ratio and angle changes of the template and the image to be detected. The template matching function module of IMAQ Vision Assistant uses a normalized cross-correlation coefficient, defined as shown in equation (2-2): (2-2) where is the average brightness of the standard template w pixels, and is the average brightness of the image to be detected f; the value of R is -1 to 1, independent of the brightness of f and w. As can be seen from equation (2-2), the cross-correlation process is a series of multiplication operations, which is quite time-consuming. If only representative parts of the image are sampled during matching, the amount of information to be processed can be greatly reduced, thereby speeding up the matching process. The IMAQ Vision Assistant pattern matching function uses two techniques to improve matching speed: non-uniform sampling and image understanding. Non-uniform sampling refers to special sampling of the detection image and the standard template to reduce spatial resolution. For example, sampling is performed every other pixel, reducing the size of the detection image and the standard template to 1/4 of their original size. Matching is performed only on the reduced image, and after matching, only the regions with high matching values ​​in the detection image are considered as regions that match the standard template. Image understanding refers to the intelligent sampling method used by the IMAQ Vision Assistant template matching function. This method considers the combination of edge pixels and matching region pixels during sampling, and also takes into account the angular rotation and scaling of the detection image. Therefore, the IMAQ Vision Assistant template matching function features high accuracy, high speed, insensitivity to image scaling and rotation, and good adaptability to brightness changes. 3. Detection of Missing Electronic Components on PCBs To avoid image blurring (noise) caused by incorrect focusing or depth of field when acquiring printed circuit board images, preprocessing—image correction—is performed first upon obtaining the PCB image to be inspected. IMAQ Vision Assistant provides the following image correction function modules: Calibrate image, Calibrate from image, and Image Correction. The Calibrate image function module corrects the image to meet the actual measurement requirements; the Calibrate from image function module requires an image containing correction information as a standard, and then compares the image to be corrected with the standard image to achieve image correction. The Image Correction function module transforms the distorted image obtained through Calibrate image or Calibrate from image correction into an image that meets the requirements. This study employed the Calibrate image function. For detecting missing electronic components on printed circuit boards, the template matching function module provided by IMAQ Vision Assistant was primarily used. First, as described in Part 2, a standard template for the component to be detected was defined, followed by parameter settings: Number of Matches of Find was set to 10; Minimum Score was set to 600; and Search for Rotated Patterns was selected with Angle Range set to 180. This not only provides information on missing components but also indicates whether component reversal occurred during the surface mount process. This information can be obtained from report files exported by IMAQ Vision Assistant, as shown in Figures 2 and 3. Figure 1 shows an image of a qualified PCB. Figures 2 and 3 show images of unqualified products. Figure 2 shows a missing component situation; in this case, the report contains no data. Figure 3 shows a reversed electronic component situation. From the report, we can see that the center position of the component matching the standard template is (358.101, 245.796), the matching value with the standard template is 753, and the angle with the standard template is 181.856 degrees, indicating that the electronic component has a reverse defect. Of course, we can also determine the skew defect of the electronic component using the angle value. 4. Conclusion The examples demonstrate that the method for detecting missing electronic components on printed circuit boards based on IMAQ Vision Assistant is feasible, effective, and easy to implement. This method has advantages such as high speed, high accuracy, strong noise resistance, insensitivity to image scaling and rotation, and insensitivity to changes in light intensity. By utilizing the powerful image processing function library of IMAQ Vision Assistant, the development cycle for detecting missing electronic components on PCBs using image processing methods is shortened, and costs are reduced. The innovation of this paper is the application of IMAQ Vision Assistant, one of NI's machine vision software programs, to the detection of missing electronic components on PCBs. References: [1] Cai Dianlin, Su Jiaxing, Lin Shijie. Application of neural network to mechanical vision inspection of surface components of printed circuit board. The 4th Precision Machinery Manufacturing Symposium, 2004 [2] Zhou Zhixian, Zheng Jinghong, Lin Shijie. Feasibility assessment of improving direct detection of defects in printed circuit board by applying subpixel interpolation method. The 4th Precision Machinery Manufacturing Symposium, 2004 [3] Zhang Guwei, Qian Dongping, Wang Jianxin, et al. Design and application of computer vision system for virtual instruments [J], Microcomputer Information, 2005, Vol. 21, 11-1: 136-138 [4] IMAQ Vision Concepts Manual [M]. USA: National Instruments, 2005 [5] Bai Changbing, Qi Chun, Song Fumin, et al. Fast visual inspection of PCB rectangular mark based on "virtual matching" [J], Special Equipment for Electronic Industry, 2005, 123: 20-24 [6] Xia Qingguan, Lu Hong, Zou Yun. Application of IMAQ-based Pattern Recognition in Part Inspection [J], Modern Manufacturing Engineering, 2005, 9: 73-75